School of Engineering and Technology Asian Institute of Technology, Thailand MODELING CARBON SINKS IN DIGITAL SPACES IN NAKHON NAYOK PROVINCE, THAILAND Examination Committee: Dr. Nitin Kumar Tripathi (Chairperson) Dr. Roland Cochard Dr. I. V. Murali Krishna Dr. Taravudh Tipdecho BY MD. AHASANUL HOQUE RS&GIS May 2011
28
Embed
School of Engineering and Technology Asian Institute of Technology, Thailand Examination Committee: Dr. Nitin Kumar Tripathi (Chairperson) Dr. Roland Cochard.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
School of Engineering and TechnologyAsian Institute of Technology, Thailand
MODELING CARBON SINKS IN DIGITAL SPACES IN NAKHON NAYOK PROVINCE, THAILAND
Examination Committee: Dr. Nitin Kumar Tripathi (Chairperson)
Dr. Roland Cochard
Dr. I. V. Murali Krishna
Dr. Taravudh Tipdecho
BY
MD. AHASANUL HOQUE
RS&GISMay 2011
Introduction
CO2 is 390 ppm (NASA, 2010).
Increasing by 1.9 ppm/yr (BBC, 2010).
Surface temperature raised by 0.74 ± 0.18 °C (IPCC, 2007)
CO2 is a prominent greenhouse gas but fundamental for photosynthesis
Statement of the Problem Kyoto protocol : reduce GHGs emission
to 5.2% below 1990 levels by 2012. => only 1 year left
About 20 countries – 80% CO2 world’s emissions.
32 countries and 10 US states - started emission trading scheme.
Countries need to know terrestrial sequestration capacity but conventional method is destructive.
Remote sensing approach - for larger area in quick turn around time and cost effective.
Objective of the Study
Developing a grid based model for CO2 absorption by different land cover classes.
DN to Reflectance ConversionDN to Reflectance Conversion
Layer StackingLayer Stacking
Subset Image by AOISubset Image by AOI
Atmospheric Correction using FLAASH Model
Atmospheric Correction using FLAASH Model
Image EnhancementContrast Stretching
-Histogram Modification
Image EnhancementContrast Stretching
-Histogram Modification
Re-projection by MRT
Re-projection by MRT
Supervised ClassificationNearest Neighborhood
Supervised ClassificationNearest Neighborhood
Accuracy AssessmentAccuracy Assessment
Ground Truth dataGround
Truth data
Land Cover ClassesLand Cover Classes
OverlayOverlay
Field Vegetation Inventory
Field Vegetation Inventory
Biomass measurement
CarbonFix Standard 3.1
Biomass measurement
CarbonFix Standard 3.1
ValidationValidation
Volume Calculation by Allometric Eqn.
Volume Calculation by Allometric Eqn.
Generate NDVI Generate NDVI
Computation of NPP using CASA
& SEABAL model
Computation of NPP using CASA
& SEABAL model
Generate NPP/ CO2 Image
Generate NPP/ CO2 Image
Masking by Land Covers
Masking by Land Covers
Above Ground Carbon stock image
by land covers
Above Ground Carbon stock image
by land covers
Output CO2
Map
Output CO2
Map
NPP = APAR . LUE = NDVI. PAR . LUE
LUE = ε0 * T1 * T2 * Λ (Field et al., 1995)
Λ = the evaporative fraction from SEBAL. (Bastiaanssen and Ali, 2001) = 0.5+ (EET/PET) (Field et al., 1995)
Where ,ε° = globally uniform maximum (2.5g/MJ) andT1 & T2 relate to plant growth regulation (acclimation) by temperature
Where,NPP=net primary productionAPAR=Absorbed Photo-synthetically Active RadiationLUE=Light Use Efficiency factorPAR= Photo-synthetically Active Radiation
CASA (Carnegie-Ames-Stanford Approach) model, to calculate NPP
Mathematical Representation of Model Algorithms
NDVI = ƒ( NIR, RED)
PAR = ƒ (K↓) (W/m2) = 0.51 for Tropical countries (Christensen and Goudriaan, 1993)
T2 = 1.185 *{1+exp (0.2 * 28 -10 – 25 )}-1 * {1+exp(-0.3* 28 -10 + 25)}-1 = 0.0016098T1 = 0.8 + (0.02*28) - 0.0005 * (28) 2 = 0.968Where PET = is potential evapotranspiration = 119.5 mm/month for Thailand (Vudhivanikh V., 1996) and 124.62 mm in December for Nakhon Nayok (Cropwat-FAO, 2006) EET = is the estimated evapotranspiration= 1.6 mm/day for Evergreen Forest in Thailand (Tanaka, N., et al, 2008) and 110.52 mm in December (RID, 2006)
LUE = 3.49 for Forest & 3.45 for Open Scrubs
Study area
Geographical coordinates
14° 12' 11" North
101° 12' 53" East . Area of 2,122 km2
60%23%
17%
Agricul-tureForest
Data Used
ASTER 15mNo. of Cells: 8774320 Area : 197422.2 ha
LANDSAT 30mNo. of Cells: 2332260 Area : 209903.4 ha
MODIS 250mNo. of Cells: 32684 Area : 204275 ha
Field Data Collection
Number of Sample Points : 31
5m
5m Tree Layer
5m
5m Tree Layer
Shrubs Layer1m
1m
Shrubs Layer1m
1m HerblayerHerblayer 15m
15m
Tree Volume Measurement
Angle Measure (%)
Ground Distance
Collector’s height1.8 m
DBH Measurement
VT = (DBH/200)2 x 3.142 x Ht / 3 (FarmForest Line, 2010)Ht = [Angle(%) x Ground Distance/100] + Collector’s Height (1.8 m)
CO2 sink measurement: Field Method
CFS was created by 60 country organizations in 1999 & released at the Climate Conference in Bali in December 2007.
According to CarbonFix Standard 3.1 (CFS), 2010
Above Ground Woody Biomass= Tree Volume x Biomass expansion Factor x Wood Density x Carbon Fraction x C to CO2 factor
Above Ground non woody biomass= Fresh biomass x Dry to wet ratio x Carbon Fraction x C to CO2 factor
Normalized Difference Vegetation Index (NDVI)
Results & Discussion
ASTER 15mNo. of pixel : 8774320
LANDSAT 30mNo. of pixel : 2332260
MODIS 250mNo. of pixel : 32684
Landcover classification images
ASTER 15mAccuracy: 89.36 %
LANDSAT 30mAccuracy: 82.98 %
MODIS 250mAccuracy: 72.34 %
Model Results of NPP
ASTER 15mSinked CO2 = 6.12 ml. ton
LANDSAT 30 mSinked CO2 = 8.83 mil. ton
MODIS 250mSinked CO2 = 10.39 mil. ton
CO2 sink = 3.25 mil. ton CO2 sink = 1.85 mil.tonCO2 sink = 3.23 mil.tonLANDSAT Primary Forest
LANDSAT Sparse Forest LANDSAT Open Scrubs
CO2 sinking by landcovers
Validation of the Model Results
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 170
0.5
1
1.5
2
Forest CO2 sink from field measurement
Sample Points
sin
ked C
O2 (
ton)
1 2 3 4 5 6 7 8 9 10 11 12 13 140
0.10.20.30.40.50.60.70.8
Open. Scrubs CO2 sink from field measurement
Sample Points
sin
ked C
O2 (
ton)
big trees
new trees
large healthy vegetation
Crop residuals
0.8 0.85 0.9 0.95 1 1.05 1.10
0.5
1
1.5
2
f(x) = − 5.0738705202 x + 5.56769036363R² = 0.754202987384012
ASTER 15m vs Forest Field Measured CO2
ASTER 15m CO2
Fie
ld C
O2
0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.40
0.5
1
1.5
2
f(x) = − 2.204561213891 x + 3.093836070613R² = 0.697496375316507
LANDSAT 30m vs Forest Field Measured CO2
LANDSAT 30m CO2Fie
ld C
O2
1 1.1 1.2 1.3 1.4 1.5 1.60
0.5
1
1.5
2
f(x) = − 0.500632270471 x + 1.37102275924R² = 0.0290934148232874
MODIS 250m vs Forest Field mesured CO2
MODIS 250m CO2
Fie
ld C
O2
R² = 0.754 R² = 0.697 R² = 0.029
Validation of the Model Results
Validation of the Model Results
0.5 0.6 0.7 0.8 0.9 10
0.050.1
0.150.2
0.250.3
0.350.4
f(x) = − 0.86232188072 x + 0.90576869276R² = 0.615759641562017
ASTER 15m vs Open Scrubs Field measured CO2
ASTER 15m CO2
Fie
ld C
O2
0.5 0.6 0.7 0.8 0.9 1 1.1 1.20
0.2
0.4
0.6
0.8
f(x) = 0.965643316214 x − 0.5092912937307R² = 0.567542565043668
LANDSAT 30m vs open Scrubs field measured CO2
LANDSAT 30m CO2F
ield
CO
20.50.60.70.80.9 1 1.11.21.31.41.5
0
0.2
0.4
0.6
0.8
f(x) = 0.3085631929346 x + 0.0006183342084R² = 0.0579838674937422
MODIS 250m vs Open Scrubs field mesured CO2
MODIS 250m CO2
Fie
ld C
O2
R² = 0.615 R² = 0. 657 R² = 0.058
Normalizing LANDSAT & MODIS to ASTER
ASTER 15m LANDSAT 30m
MODIS 250m0.00
2.00
4.00
6.00
8.00
10.00
12.00
6.12
8.33
10.39
sinked CO2
Images
sin
ekd C
o2 (
million t
on)
Converting value to Same Unit area 15m x 15m multiplied
By 4 to Landsat & 277 to Modis
ASTER 15m LANDSAT 30m
MODIS 250m
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.006.12
2.08
0.04
sinked CO2
Images
sin
ked C
O2 (
million t
on)
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.80
0.10.20.30.40.50.60.70.80.9
f(x) = 0.804694551529963 x + 0.247417541477452R² = 0.739154576842278
NDVILinear (NDVI)
MODIS 250m NDVI
LA
ND
SA
T 3
0m
ND
VI
LANDSAT 30m MODIS 250m0
2
4
6
8
10
12
8.33 8.33
2.06
ScaledActual
Images
sin
ked C
O2 (
million t
on)
LANDSAT 30m & MODIS 250m Image acquisition date is same LANDSAT 30m NDVI & MODIS 250m NDVI 41 points correlation is 0.739
Therefore , assuming LANDSAT 30m result is actual CO2 sink value
Actual CO2 sink = 0.801 x MODIS 250m sink (ton)
Downscaling MODIS 250m to MODIS 30m
MODIS 30m NDVI MODIS 30m Land Cover Classes No. of pixel = 2332116Sinked CO2 =2.66 mil. ton
Downscaling MODIS 250m to MODIS 15m
No. of Cells = 9515056Sinked CO2 = 10.85 mil. ton
MODIS 15m NDVI MODIS 15m Land Cover Classes
Comparison among downscaled image and other images sink
0.00
2.00
4.00
6.00
8.00
10.00
12.00
6.12
2.08
0.04
2.66
10.85sinked ...
images
sink
ed C
O2
in m
illio
n to
n
ASTER 1
5m
LANDSAT.
..
MODIS
...
ds_M
ODI..
.
ds_M
ODI..
.0
2
4
6
8
10
12
6.12
8.33
10.39 10.62 10.85sinked ...
Images
sin
ked C
O2 in m
illion t
on
Pixel based Value of each images sink
Normalized sink value of each image
MODIS 250m to LANDSAT 30m sink
ASTER
15m
MOD
IS 1
5m
LAND
SAT
30m
MOD
IS 3
0m0
2
4
6
8
1012
6.12 6.128.33 8.33
4.732.29
scaledactual
Images
Sin
ked
Co2
( m
illio
n t
on)
ASTER 15m, LANDSAT 30m & downscaled MODIS sink comparison
LANDSAT 30m sink = 0.79 x downscaled MODIS 30m sink
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.80
0.10.20.30.40.50.60.70.80.9
f(x) = 0.862208280173626 x + 0.201038337478708R² = 0.718970962613201
NDVILinear (NDVI)
MODIS 30m NDVI
LA
ND
SA
T 3
0m
ND
VI
NDVI correlation between LANDSAT 30m and MODIS 30m
R² = 0.719
Conclusion CASA algorithms can give the precise and accurate result if SEBAL data are of
same time The finer the satellite resolution the more accurate CO2 result.
Medium resolution LANDSAT provided net CO2 sink stock estimate as 8.33 million tonnes of CO2 in Nakhon Nayok.
Scale factor 0.801 for MODIS 250m for achieving the amount of LANDSAT 30m sequestered CO2.
A general average stock can be assumed to meet Kyoto Protocol or CDM target of carbon quantities.
MODIS data can be used to find terrestrial sink accurately.
Recommendations
Proposed methodology with MODIS can be used for regional, or global scale carbon sequestration measuring and monitoring tool.
Research to measure the CO2 emission data of the study area & required afforestation for remaining CO2.
MODIS 250m CO2 Map
With deep gratitude to Govt. of Japan for supporting to study in AIT.